Andreas Hansen

Receding horizon sliding control for linear and nonlinear systems

Sliding mode control (SMC) is one of the few controller design methodologies that can be applied to highly nonlinear and uncertain systems. In most mechanical applications, a smoothed version of SMC that we call “sliding control” is employed to keep the system trajectories close to but not necessarily on a stable differential/difference manifold. In this paper, we propose an extension to the sliding control algorithm that includes a receding horizon approach. This allows the designer to incorporate information about future desired output trajectories as well as knowledge of future disturbances. In addition, it allows the designer to include bounds on the inputs and states of the system. We refer to this control methodology as receding horizon sliding control (RHSC) and evaluate its effectiveness using two example applications.

Learning a deep neural net policy for end-to-end control of autonomous vehicles

Deep neural networks are frequently used for computer vision, speech recognition and text processing. The reason is their ability to regress highly nonlinear functions. We present an end-to-end controller for steering autonomous vehicles based on a convolutional neural network (CNN). The deployed framework does not require explicit hand-engineered algorithms for lane detection, object detection or path planning. The trained neural net directly maps pixel data from a front-facing camera to steering commands and does not require any other sensors. We compare the controller performance with the steering behavior of a human driver.

Algorithmic performance of receding horizon sliding control for engine emission reduction

This paper presents a performance assessment of the recently developed receding horizon sliding control (RHSC) algorithm. The RHSC strategy is extended to multi-input multi-output systems and will be evaluated using a model predictive control (MPC) approach as a benchmark. The focus of this investigation is on comparing performance in computational logistics and output-tracking between the two control strategies. We show that RHSC can improve the level of tracking accuracy when compared to classical MPC. At the same time, it is on average more computationally efficient. The case study for evaluation is a nonlinear automotive engine cold-start emissions system.

A Computationally Efficient Predictive Controller for Lane Keeping of Semi-Autonomous Vehicles

Model predictive control (MPC) is a popular technique for the development of active safety systems. However, its high computational cost prevents it from being implemented on lower-cost hardware. This paper presents a computationally efficient predictive controller for lane keeping assistance systems. The controller shares control with the driver, and applies a correction steering when there is a potential lane departure. Using the explicit feedback MPC, a multi-parametric nonlinear programming problem with a human-in-the-loop model and safety constraints is formulated. The cost function is chosen as the difference between the linear state feedback function to be determined and the resultant optimal control sequence of the MPC problem solved off-line given the current state. The piecewise linear feedback function is obtained by solving the parametric programming with an approximation approach. The effectiveness of the controller is evaluated through numerical simulations.Copyright

A receding horizon sliding control approach for electric powertrains with backlash and flexible half-shafts

ABSTRACT Active control of electric powertrains is challenging, due to the fact that backlash and structural flexibility in transmission components can cause severe performance degradation or even instability of the control system. Furthermore, high impact forces in transmissions reduce driving comfort and possibly lead to damage of the mechanical elements in contact. In this paper, a nonlinear electric powertrain is modelled as a piecewise affine (PWA) system. The novel receding horizon sliding control (RHSC) idea is extended to constrained PWA systems and utilised to systematically address the active control problem for electric powertrains. Simulations are conducted in Matlab/Simulink in conjunction with the high fidelity Carsim software. RHSC shows superior jerk suppression and target wheel speed tracking performance as well as reduced computational cost over classical model predictive control (MPC). This indicates the newly proposed RHSC is an effective method to address the active control problem for electric powertrains.

Nonlinear control design within the high level modeling framework for an engine cold start scenario

The high level modeling (HLM) paradigm is a rapid modeling method that is being developed by the Toyota Motor Corporation. Emphasizing the use of physical conservation laws, it streamlines the model finding process for control applications. While recent works show the successful application of the high level modeling approach to physical systems of different complexity levels, this work focuses on control design for HLM plant descriptions. Based on the exemplary scenario of automotive cold start emission reduction, this work first explains how model simplification is integrated in the controller development process. Subsequently, a nonlinear sliding controller for reducing harmful emissions during the cold start phase is derived based on a HLM description of the underlying dynamics. Thereby, ideas from dynamic surface control are used to conveniently deal with the computer-generated system equations. The obtained control laws are evaluated using numerical simulations of the closed-loop system.

Validation of uncertainty estimation in engine cold start

The engine cold start period produces the majority of harmful hydrocarbon emissions during engine operation, and, therefore, the reduction of such emissions is key to maintaining ultra low emission vehicle standards. Emissions reduction can be achieved by the design and implementation of controllers with effective tracking performance and estimation of uncertain parameters in the system. Effective tracking performance can be achieved by driving down error in the tracking of desired engine state trajectories. Estimation of uncertainty can be achieved through the use of an adaptive controller. In this work, an adaptive sliding controller is designed in order to achieve both goals, and implemented on an engine test cell. Experimental results show the reduction of the tracking error as well as estimation of model uncertainty in the engine test cell.

Adaptive engine cold start emission control

Reduction of hydrocarbon emissions during the engine cold start process is a major design and control goal in the automotive industry in recent years, with considerable impact on not only the vehicles fuel economy, but the environment as well. The key to producing the most emission efficient powertrain system is the ability to drive the engine variables to operate under certain behavioral parameters, that is, to drive the engine states to follow ideal desired trajectories. Therefore, the control target of driving down tracking error to produce ideal engine behavior is an important consideration in the control design process. Due to the highly nonlinear and transient nature of the engine cold start process, the choice was made in this work to employ a scalar sliding control technique. Additionally, in order to mitigate possible model uncertainty in real time, an adaptation update algorithm was incorporated. The combined algorithm was implemented on an engine test cell and experimentally validated. The work was inspired by the Verification & Validation procedure used in standard industry practice to reduce errors and uncertainties early on in the control design phase so as to produce desired engine behavior as soon as possible.

Handling of Partially Filled Tank Containers by Means of Cranes

Tank containers are widely used to transport a variety of liquid goods such as food products, oil, and different kinds of fuel including liquefied natural gas. Due to the unpredictable dynamic behavior of partially filled tank containers, regulations limit the containers to be either almost full (> 80%) or almost empty (< 20%), when handled by cranes. In order to provide arguments to ease these restrictions, the system is analyzed and control methods for assisting the crane operator are proposed. We deduce a very accurate and computationally favorable mathematical description of the coupled crane and fluid dynamics. The fluid is modeled by a potential flow approach resulting in a low dimensional approximation of the liquid dynamics. Coupling the fluid dynamic model with the load system model of a container crane leads to a nonlinear formulation of the overall system. The state estimation algorithm exclusively relies on the measured rope forces as well as the known motion parameters of the trolley and the rope winches. A nonlinear state feedback controller based on sliding modes for underactuated systems provides a stabilizing control signal for the system. Experimental results for validation of the model, the observer, and the control design are included.Copyright